Update README.md

README.md

@@ -11,12 +11,27 @@ language:
   - zh
   - ja
 pipeline_tag: text-generation
-base_model: anthracite-org/magnum-12b-v2
+base_model: anthracite-org/magnum-v2.5-12b-kto
 tags:
 - chat
 ---
+# magnum-v2.5-12b-kto-exl2
+Original model: [magnum-v2.5-12b-kto](https://huggingface.co/anthracite-org/magnum-v2.5-12b-kto)  
+Creator: [anthracite-org](https://huggingface.co/anthracite-org)
 
+## Quants
+[4bpw h6 (main)](https://huggingface.co/cgus/magnum-v2.5-12b-kto-exl2/tree/main)  
+[4.5bpw h6](https://huggingface.co/cgus/magnum-v2.5-12b-kto-exl2/tree/4.5bpw-h6)  
+[5bpw h6](https://huggingface.co/cgus/magnum-v2.5-12b-kto-exl2/tree/5bpw-h6)  
+[6bpw h6](https://huggingface.co/cgus/magnum-v2.5-12b-kto-exl2/tree/6bpw-h6)  
+[8bpw h8](https://huggingface.co/cgus/magnum-v2.5-12b-kto-exl2/tree/8bpw-h8)  
 
+## Quantization notes
+Made with exllamav2 0.2.2 with the default dataset.  
+These quants are for RTX cards on Windows/Linux or AMD on Linux.  
+Use with Text-Generation-WebUI, TabbyAPI, etc.  
+
+# Original model card
 ![image/png](https://cdn-uploads.huggingface.co/production/uploads/658a46cbfb9c2bdfae75b3a6/sWYs3iHkn36lw6FT_Y7nn.png)
 
 v2.5 KTO is an experimental release; we are testing a hybrid reinforcement learning strategy of KTO + DPOP, using rejected data sampled from the original model as "rejected". For "chosen", we use data from the original finetuning dataset as "chosen".